The Discussions on Traffic Reform with the Lindens

I don't care if it just uses the word salad or both that and the description for indexing. We definitely need to be able to have a description that isn't just word salad. I agree with Sling. If keywords are the only avenue available to people to manipulate search results that's a major improvement over what's going on now with traffic and picks.

Another benefit of having the searches via standard web search engines would be that any classic keyword spamming moves would fall foul of the measures implemented by the engines to prevent such things.
They're the professionals, and vastly more able to deal with that than would be a LL developer reinventing search from the ground up on his/her own.

How about if the words and descriptions of objects are used for the search index, then the description fields can contain plain English, or of course whatever your native language is.

This is of course a bit of an issue for places that don't sell objects.

Agreed. I wonder how much load it would remove from the servers if they didn't have to poll all Picks and Traffic?

What if it isn't the worst place in SL for what it does, and what if worse places are above it? The thing is, it's *relevant* for what it does.

oooo - that's patently untrue. Have you been to my place? People complain about traffic out of principle, and a misguided idea of their effect on load.

I don't agree with that. Imo, store owners are as good anyone else to determine what's good and what's not good. But that's not the point. The thing is that link acquisition is a standard way of dealing with search engines, which the engines themselves suggest and recommend. We are talking about a search engine here, and nota perfect search system. No search engine is perfect - they can't be - and the SL one can't be perfect either. For instance, suppose that, by the relevancy algorithm, crap places rank at the top and great places rank low down where people won't find them. Do you suggest that it should be left that way even though it's bad for users? It's either leave things that way, or take steps to move the great places up the rankings, and if you do that, how many steps? Where do you draw the line?

Alright, but when the wording of a Pick on someone's profile is changed for whatever reason, it no longer helps with the relevant rankings that it did before the change, so those Picks aren't counted for relevancy anyway.

True.

Yes they can and do. It's an essential part of seo.

For the Places search, it couldn't happen, because it's an extremely basic search and doesn't deal with plurals etc. Using your example, the phrase "female skins" isn't likely to be in the name or description of an object on the parcel, but it's the phrase that people will search on, so it'll be in the parcel's title and/or description. A match wouldn't occur, so you'd have owners 'arranging' their objects' names and descriptions, and we're back to square one. Not only that, but many places simply wouldn't be able to have matching phrases in the content and the parcel description for all their varying items, due to the limited number of characters allowed, so they simply would show in search for some of their stuff, and they should. Also the user wouldn't be able to find them for the missing things, and they should be able to.

It couldn't happen in the All search because LL doesn't have control of the algo, and it doesn't do that. If it did, we'd still be back to square one again.

Well, a) most do, since they often are built around showcasing the produces they make, even when designed for more than that, and b) this matter only if you looking for shopping. Its a lot easier to find, for example, a state park, than it is to find the best place to buy toothpicks. The reason is that when you want to look for a state park, you are looking at size, what sort it is, and other factors, like if its 50 miles of desert, or 50 miles of pine trees, than if it sells post cards with pictures of squirrels on them. There is no reason why a well designed system can't do that.

See, I personally think that the traffic problem comes down to *two* issues:

1. Right now you can't find any place to go, unless its word of mouth, you have a *fairly* clear idea what it is you are looking for, or you pick stuff at random from some stupid traffic ranked list, and hope its interesting.

2. If people can't find any place to go and do interesting things, they are going to go to where the traffic is, which will drop them in mega-malls, badly made dance clubs, and BDSM locations (mostly). Your not going to get "much" better results turning off "mature content", since half the "good" areas are in that category, due to what is allowed in them, despite the fact that its ***not*** supposed to be their primary draw.

Fix it so people can find real content, and you fix half the traffic issue right off the bat. You still can't do anything about bots, but if someone knows that everything with traffic over 200k is a bot zone, or can figure that out fairly easy, they are going to just automatically look elsewhere. As things stand, you can't get legit traffic to good content easily, so your stuck hoping that the people looking will scroll down 5 pages, before clicking a link.

Put simply, let them "find" what they want to look for, then if they care, also show them the traffic for that place, then let them decide if they *trust* the numbers. Don't feed everything through the numbers, give them no way to properly search, then presume that this is OK, because newbies can *ask* where good places are. I have been to dozens of those "good places", and probably hit less than 1% of the ones available. Most of the time I see less than 2-3 people in them. Other places, are pure junk, but have 30 bots, and 20 real people, all the new ones probably wondering, "Gosh! Does all of SL suck like this?"

You don't fix traffic by changing traffic, you fix it by helping people find a **better** way to get places in the first place. Traffic 'should' fix itself after that (or at least way more than trying to come up with some new gamable system is going to).

Still a bad idea. We don't need something that just tells you it sells skins. Personally, I want to know *what* skins. If Joe Somebody makes better ones than Joe Nobody, I want to know which parcels sell the Somebody's, not which ones just sell them. Its not good enough, and its being mangled by traffic information, which is useless without context. And that is what is missing, no **context** for what is really there, what else is offered, like if it happens to sell clown hats ***because*** its an amusement park. Sure, those keywords are probably going to be in there, but someone looking for clown hats is probably going to get 5 pages of people that sell clown hats as an item, along with whips and chains, or something, and the poor guy selling "only" clown hats, and a few other odds and ends, to help keep his park open, is basically screwed. Its 180 degrees backwards from the way it should work. Its like going to pawn shops to buy yourself a new TV, because they put a plaque on the front door saying, "We have TVs, as well as knives, glass eyes, wigs, wedding rings, appliances, etc., etc., etc.", and 50 people a day come through the door to sell the family heirlooms for a quick buck. Its **not** where anyone not looking to make a quick buck, i.e. campers, is going to even want to know about, never mind look for anything.

The problem with the clown hat analogy is that the guy selling ONLY clown hats to support his circus sim is obviously not wording or subdividing his parcels correctly.

If, say, he has a circus sim, he should list that main parcel as "Circus Rides Games Animals Clowns blah blah" and then subdivide his shop out, which is presumably near the main circus parcel, and list it as "Clown Hats Circus Costumes Outfits Masquerade etc." That way both would show up at the top of Search All for their respective areas, and he would get basically the same consumers for both. In other words, the shop parcel would get casual spillover shoppers from the circus, and the circus would get spillover guests from the shop. Win-win either way.

Let's not forget something important. People who pay for picks had better have the sales to back it up. I mean, let's be real.

Just because someone visits a store doesn't mean that the visit translates into a sale unless they have something the visitor wants.

Someone that pays $100 USD/wk for picks camping likely has the sales to back it up and it's not necessarily a bad thing if they have things that people want, and those visits translate into sales. If they don't have the sales to back it up, guess what? That entity typically goes away. The shop on Skin Oasis doesn't have to pay anyone for its bot farm. But guess what happens if they start paying people to advertise, but don't get the requisite sales? They are going to go away.

The system will take care of itself.

That's what they said about camping. Then the traffic bots came.

Well, yeah, but point is, that only works if you are making it a shop primarily. As things stand, you are *more* likely to find it as a shop than as a place of entertainment, no matter "what" terms you use, or how you do it. And, there is no way to say, "Ok, so a shop sells costumes there, is there *any* other reason to go to that region, when I can just pop over to 'Sam's Mega Costume Shop and Camping Bar', and get it there. It would be nice of searches for items let you check what else was in the region, what the region itself was designated as, or even which other shops where there. I happen to use a tail, hair and some stuff from one avatar with another. Maybe, if I wanted to upgrade, I would like to find a place with "both" of those merchants in it, to see if I want to mix and match, or just get one of them. There isn't enough cross referencing to draw people in to "other" content in your areas, so you are stuck with them maybe not even knowing about it, unless they happen to walk out of the shop to look (which on some areas they may not do.

For example, there is one place that is like 5-6 floors of shops in a castle, which is "connected" to 3-4 RP zones. There where 3-4 people in the shop, but unless you went out the door of the castle, then flew down to one of the RP areas, you would never know. Bad design? Maybe, but good design in this case would have meant putting the shop in the town square of some village, where it would have "detracted" from the content, instead of enhancing it.

All they need is a way to get their "alt army" to appear in search, which is possible and has negliable costs associated with it, at which point they can play the "pick farm" game instead of the "bot farm" game.

It will not go away, it will merely transform into something different. Any system that attributes the same value to disposable alts as it does to individuals and is predictably "bribeable" will ultimately end up heavily gamed.

Sale will probably increase seeing in the "pick bot" game they don't need to login the AV's grinding the sim to a halt, but then again, they'll probably do it anyway.....

As far as I have seen, there hasn't been a way to do that. From what I understand, the tracking is a little more sophisticated than just the name to email match. Correct me if I'm wrong, Phil, but have you been able to get alts into search yet?

The bot has to have a HTML page or it won't count.

What's the minimum that needs to be spent in order to get an avatar a web page in the current New Search?

If it's so that LL have become good at detecting alts and preventing them from influencing the new web-based Search, then they could equally well flag such avatars internally and keep them from counting towards the current Traffic.
The alts would still be able to inflate concurrency ( which is said to be something that LL desire ) but they wouldn't subvert the current or future Search ranking.

But then, if the game was up for traffic bots and camping bots, the game would also be up for artificially inflated concurrency.

Check my sig.
Given a choice between high concurrency and usability, LL will go for concurrency.
They want Search to be more meaningful, but at the same time they don't want to kill the incentive for people to create armies of accounts. That's never going to end well.

No idea and it's not RL spending that's for sure because I have two alts, one who has a page, one who doesn't, all tied to the same email address, neither had payment info on file.

The major difference for my alts has been that one is rather active, spends Linden dollars on classifieds etc. and the other is just used for test purposes so logs in nowhere near as much.

Anything that links back to the avie's profile is a likely candidate: create an event (free), create a public group (L$100), or own a parcel that's listed in search (L$30).

I think people tried with groups in the "NPIOF aren't listed in search" and that worked, not sure if anyone tried with an event.

I agree with that main point: if LL is willing to cut alts from the new search then it would be better to simply dump the new search and apply the same screening and keep bots/alts from counting towards traffic which is quite easy to do as suggested in the "Traffic reform" thread.

Traffic is also an inherently better metric because it's dynamic rather than static, and doesn't suffer from historical pollution since it's recalculated from scratch day-to-day (people create picks for friends, places move and picks don't get updated, accounts fall into disuse but their picks still count, etc) and the old search doesn't suffer from new ways to cheat like "google bombing" of keywords.

The old search is worse at matching ("cat" doesn't match "cat/cats/cat's", but that's easy enough to remedy.

Ah a system charge, that would explain it.

Classifieds used to be listed in the new search and they contained both a link to the place they were set to, and the avie that created them which likely pulled your alt's profile into the new search.

(Changed my previous post to add that owning a parcel that's listed in search creates a link back to the owning avie's profile as well, which also makes it a candidate for having a profile added to the new search)

Of course the reasons for one alt not having a page could include
- The indexing was done on a Tuesday
- The alt has non-prim hair
- The alt has a strange name
- The indexing was done on Friday
- <a grillion other borky reasons>

It would be nice of LL came out in the open about any grouping of accounts by suspected type for ranking purposes. Right now, given the history and the newness, I'd be more inclined to put the absence of a web page down to brokenness rather than to conspiracy.

Failing openness by LL, one could tour a large number of bot camps and check samples of bot population for presence of web pages.